JP2013025466A - Image processing device, image processing system and image processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing device, image processing system, and image processing program that are able to generate a highly accurate microscope image at a high speed.SOLUTION: The image processing device according to the present technique comprises an image capturing part, tile area setting part, temporary tile area setting part, temporary tile image formation part, noise removal part, and tile image formation part. The image capturing part captures a scan image. The tile area setting part divides a scan image into a plurality of tile areas. The temporary tile area setting part has a tile area in a scan image and sets a temporary tile area larger than the tile area. The temporary tile image formation part extracts a scan image from each temporary tile area and generates a temporary tile image. The noise removal part removes noise from the temporary tile image. The tile image formation part extracts a temporary tile image from each tile area and generates a tile image.

Description

  The present technology relates to an image processing apparatus, an image processing system, and an image processing program related to tile division processing of a scan image captured through a microscope in pathological diagnosis.

  In the field of pathological diagnosis and the like, a process called tiling (tile division) is often executed in order to process a high-resolution microscope image. When the observation object is imaged at a high resolution (high magnification), the imaging range of one imaging is narrowed. However, different areas of the observation object are sequentially imaged, and a plurality of captured images (scan images) are connected (stitch). To obtain one large high-resolution image.

  In tiling, a scan image of an observation object is divided into tiles of a predetermined size (for example, 256 pixels × 256 pixels) and stored in an HDD (Hard Disk Drive) or the like. As a result, the memory in which the scan image is read can be released and the next scan image can be acquired. For example, Patent Document 1 describes that “images are tiled” acquired by scanning.

JP 2009-37250 A (paragraph [0036], FIG. 16)

  Here, in a microscope image, in particular, a fluorescence microscope image captured using a fluorescence microscope, it is common to remove noise contained in the image by a mathematical method (wavelet transform or the like). However, when noise removal processing is performed on the scanned image, there is a problem that it is difficult to generate a microscope image with high speed and high accuracy due to a phenomenon caused by tiling.

  In view of the circumstances as described above, an object of the present technology is to provide an image processing apparatus, an image processing system, and an image processing system capable of generating a microscope image with high speed and high accuracy.

To achieve the above object, an image processing apparatus according to an embodiment of the present technology includes an image acquisition unit, a tile region setting unit, a temporary tile region setting unit, a temporary tile image generation unit, a noise removal unit, a tile, An image generation unit.
The image acquisition unit acquires a scanned image.
The tile area setting unit divides the scan image into a plurality of tile areas.
The temporary tile area setting unit sets a temporary tile area that includes the tile area and is larger than the tile area in the scanned image.
The temporary tile image generation unit generates the temporary tile image by extracting the scan image for each temporary tile area.
The noise removal unit performs noise removal processing on the temporary tile image.
The tile image generation unit generates the tile image by extracting the temporary tile image for each tile region.

  According to this configuration, when the temporary tile image generation unit generates the temporary tile image, the memory of the image processing apparatus is released and the next scan image can be acquired. Since noise removal processing is not executed in the meantime, the memory can be released early. Therefore, in the image processing apparatus, the image acquisition unit can quickly acquire the next scan image. In addition, when the noise removing unit performs noise removal processing on the temporary tile image, a “discontinuous region” is generated at the periphery of the temporary tile image due to lack of information on continuous pixels. On the other hand, the tile image generation unit uses the tile area of the temporary tile image subjected to the noise removal process as a tile image, thereby preventing the tile image from including a discontinuous area. Therefore, in the present image processing apparatus, when tile images are arranged and displayed, it is possible to prevent the images from becoming discontinuous at the boundary between adjacent tile images.

  The image acquisition unit acquires a plurality of continuous scan images, the image processing apparatus further includes a stitching unit that calculates a relative position of the plurality of scan images, and the tile region setting unit includes: The scanned image may be partitioned into a plurality of tile areas based on the relative position.

  According to this configuration, when the tile area setting unit sets a tile area for a specific scan image, the tile area can be set across the continuous scan images. However, it is possible to generate a temporary tile image using continuous scan images. Thereby, even when a plurality of scan images are captured with respect to one observation object, it is possible to generate tile images that are continuous between the plurality of scan images.

  The scan image may be a fluorescence microscope image captured through a fluorescence microscope.

  In the fluorescence microscope image, the brightness of the observation object is small in principle, and therefore it is necessary to execute exposure by the image sensor for a long time. For this reason, noise is easily generated in the fluorescence microscope image, and noise removal processing by the image processing apparatus of the present technology is particularly effective.

In order to achieve the above object, an image processing system according to an embodiment of the present technology includes an imaging device and an image processing device.
The imaging device captures a scan image of the observation object.
The image processing apparatus includes an image acquisition unit that acquires the scan image, a tile region setting unit that partitions the scan image into a plurality of tile regions, and a temporary image that includes the tile region in the scan image and is larger than the tile region. A temporary tile area setting unit that sets a tile area, a temporary tile image generation unit that generates a temporary tile image by extracting the scan image for each temporary tile area, and noise removal that performs noise removal processing on the temporary tile image And a tile image generation unit that generates the tile image by extracting the temporary tile image for each tile region.

  The imaging device may image the observation object via a fluorescence microscope.

In order to achieve the above object, an image processing program according to an aspect of the present technology includes an image acquisition unit, a tile region setting unit, a temporary tile region setting unit, a temporary tile image generation unit, a noise removal unit, a tile, The computer functions as an image generation unit.
The image acquisition unit acquires a scanned image.
The tile area setting unit divides the scan image into a plurality of tile areas.
The temporary tile area setting unit sets a temporary tile area that includes the tile area and is larger than the tile area in the scanned image.
The temporary tile image generation unit generates the temporary tile image by extracting the scan image for each temporary tile area.
The noise removal unit performs noise removal processing on the temporary tile image.
The tile image generation unit generates the tile image by extracting the temporary tile image for each tile region.

  As described above, according to the present technology, it is possible to provide an image processing apparatus, an image processing system, and an image processing system capable of generating a microscope image at high speed and with high accuracy.

1 is a block diagram illustrating a functional configuration of an image processing system according to an embodiment of the present technology. It is a conceptual diagram for demonstrating the imaging of the imaging target object by the imaging device which comprises the image processing system. It is an example of the scanned image which the image acquisition part of the image processing apparatus which comprises the image processing system acquires. It is a conceptual diagram which shows the stitching by the stitching part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the tile area | region set to the scan image by the tile area | region setting part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the temporary tile area | region set to the scan image by the temporary tile area | region setting part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the temporary tile image produced | generated by the temporary tile image production | generation part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the noise removal by the noise removal part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the tile image produced | generated by the tile image production | generation part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the arranged tile image displayed on the display part of the image processing apparatus which comprises the image processing system. It is a conceptual diagram which shows the noise removal method which concerns on the comparative example 1. FIG. It is a conceptual diagram which shows the noise removal method which concerns on the comparative example 2. It is a mimetic diagram showing composition of a fluorescence microscope which is a specific example of an imaging device which constitutes an image processing system concerning an embodiment of this art. It is a schematic diagram which shows the structure of the computer for microscopes which is a specific example of the image processing apparatus which comprises the image processing system. It is each image which shows the effect of the image processing system. It is a graph of the luminance distribution of each image which shows the effect of the image processing system. It is a graph of the luminance distribution of each image which shows the effect of the image processing system. It is a graph of the luminance distribution of each image which shows the effect of the image processing system.

  Hereinafter, embodiments according to the present technology will be described with reference to the drawings.

<Functional configuration of image processing system>
A functional configuration of the image processing system according to the present embodiment will be described.

  FIG. 1 is a block diagram illustrating a functional configuration of an image processing system 1 according to an embodiment of the present technology. As illustrated in FIG. 1, the image processing system 1 includes an image processing device 10 and an imaging device 20, and the imaging device 20 is connected to the image processing device 10. The image processing apparatus 10 is, for example, a microscope computer, and the imaging apparatus 20 is, for example, a fluorescence microscope imaging apparatus.

  The image processing apparatus 10 includes an image acquisition unit 11, a stitching unit 12, a tile region setting unit 13, a temporary tile region setting unit 14, a temporary tile image generation unit 15, a noise removal unit 16, a tile image generation unit 17, and a storage unit 18. And a display unit 19. The image acquisition unit 11, the stitching unit 12, the tile area setting unit 13, the temporary tile area setting unit 14, and the temporary tile image generation unit 15 are sequentially connected in this order, followed by the temporary tile image generation unit 15, the storage unit 18, and the noise. The removal unit 16, the tile image generation unit 17, and the storage unit 18 are connected in this order. The display unit 19 is connected to the storage unit 18.

  The imaging device 20 captures a scan image of the imaging object. FIG. 2 is a conceptual diagram for explaining the imaging of the imaging target P by the imaging device 20. As shown in the figure, on the imaging object P, an imaging range S1 that is imaged by one imaging (one shot) of the imaging device 20, an imaging range S2 that is imaged by the next imaging, and further by the next imaging An imaging range S3 to be imaged is shown.

  When the imaging device 20 is a microscope imaging device, the imaging range of one imaging varies depending on the magnification, but it does not reach the entire imaging target P and may be a partial region of the imaging target P. It is normal. Accordingly, in order to capture a high-magnification image of the entire imaging target P, it is necessary to capture the image while shifting the imaging range. Note that the imaging range can be determined by area detection processing from thumbnail images obtained by imaging the entire imaging target P at a low magnification.

  Hereinafter, an image captured by one imaging by the imaging device 20 is referred to as a “scanned image”. The number of pixels of the scan image is the number of pixels of the image sensor of the imaging device 20, and can be set to, for example, 24M pixels (24 million pixels). The imaging apparatus 20 outputs these scan images to the image acquisition unit 11 of the image processing apparatus 10 sequentially for each imaging.

  The image acquisition unit 11 acquires the scan image output from the imaging device 20 as described above. FIG. 3 is a diagram illustrating an example of a scanned image acquired by the image acquisition unit 11. 3A is a scan image G1 captured in the imaging range S1, FIG. 3B is a scan image G2 captured in the imaging range S2, and FIG. 3C is a scan image captured in the imaging range S3. G3 is shown. The image acquisition unit 11 sequentially supplies these scan images to the stitching unit 12.

  The stitching unit 12 “stitches” the scanned image supplied from the image acquisition unit 11. FIG. 4 is a conceptual diagram showing stitching. As shown in the figure, stitching means calculating relative positions of scan images captured separately. The stitching unit 12 can calculate the relative position of the scan image using, for example, a correlation function of luminance values for each pixel of the scan image.

  For example, as shown in FIG. 4A, the stitching unit 12 calculates the relative position of the scan image G2 with respect to the scan image G1 when the scan image G1 and the scan image G2 are supplied from the image acquisition unit 11. Subsequently, as shown in FIG. 4B, the stitching unit 12 calculates the relative position of the scan image G3 with respect to the scan image G2 when the scan image G3 is supplied. Thereby, the stitching unit 12 can calculate the relative position of the scan image G3 from the scan image G1.

The stitching unit 12 does not generate an image obtained by actually connecting scan images, but calculates a relative position (for example, an offset value) between the scan images. The stitching unit 12 supplies the calculated relative position to the tile area setting unit 13. In the following description, one of the scan images G1 to G3 is simply referred to as a scan image G.

  The tile area setting unit 13 divides the scan image G into a plurality of “tile areas”. FIG. 5 is a conceptual diagram showing the tile region R1 set in the scan image G. The tile area R1 is an area that becomes a division unit when the tile image generation unit 17 described later generates a “tile image” by dividing the scan image G. The tile region R1 can have an arbitrary size, but a size of 256 pixels × 256 pixels is preferable for convenience of image capacity and arithmetic processing.

  Although the tile area R1 shown in FIG. 5A coincides with the periphery of the scan image G, for example, the tile area R1 may protrude from the scan image G as shown in FIG. In this case, there is no problem because the tile image generation unit 17 uses a part of another scan image based on the relative position calculated by the stitching unit 12. The tile area setting unit 13 supplies the tile area R1 to the temporary tile area setting unit 14 together with the scan image G.

  The temporary tile area setting unit 14 sets a “temporary tile area” in the scanned image. FIG. 6 is a conceptual diagram showing the temporary tile region R2 set in the scan image G. The temporary tile area R2 may be an area that includes at least one tile area R1 and is larger than the tile area R1.

  As shown in FIG. 6A, the temporary tile region R2 can be a region having a predetermined width larger than each tile region R1. Specifically, when the size of the tile area R1 is 256 pixels × 256 pixels, the temporary tile area R2 can be 288 pixels × 288 pixels (that is, the predetermined width is 16 pixels). The predetermined width is not particularly limited, but a pixel having a multiple of 8 is desirable in order to reduce the burden of calculation processing in noise removal described later. Further, as shown in FIG. 6B, the temporary tile region R2 can be a region having a predetermined width larger than a plurality of adjacent (for example, four or nine) tile regions R1.

  As described above, since the tile area R1 is an area that partitions the scan image G, each tile area R1 does not overlap with the adjacent tile area R1. However, since the temporary tile region R2 is a region having a predetermined width larger than the tile region R1, the temporary tile region R2 overlaps with the adjacent temporary tile region R2. The temporary tile area setting unit 14 supplies the temporary tile area R2 to the temporary tile image generation unit 15 together with the tile area R1 and the scan image G.

  The temporary tile image generation unit 15 generates a “temporary tile image” from the scan image G. FIG. 7 is a conceptual diagram showing the temporary tile image X. As shown in the figure, the temporary tile image generation unit 15 extracts a scan image for each temporary tile region R2 in the scan image G and generates a temporary tile image X. As described above, since the temporary tile area R2 overlaps with the adjacent temporary tile area R2, the image areas included in the two or more temporary tile areas R2 are included in each tile image X.

  The temporary tile image generation unit 15 operates as follows when the tile region R1 (and the temporary tile region R2) set by the tile region setting unit 13 protrudes from the scan image G (see FIG. 5B). . That is, the temporary tile image generation unit 15 extracts an image region that should be included in the temporary tile region R2 from the adjacent scan images G based on the relative positions between the scan images G calculated by the stitching unit 12, and A tile image X is generated. The temporary tile image generation unit 15 supplies the generated temporary tile image X to the storage unit 18.

  The storage unit 18 stores the temporary tile image X supplied from the temporary tile image generation unit 15. At this time, the memory used for the processing of the temporary tile image generation unit 15 from the image acquisition unit 11 is released. That is, the image acquisition unit 11 to the temporary tile image generation unit 15 can execute the above-described processing on a new scan image.

  The noise removal unit 16 reads each temporary tile image X stored in the storage unit 18 and executes noise removal processing. Noise removal can be performed by various algorithms such as wavelet transform and high frequency component cut. Specifically, using wavelet toolbox (registered trademark: the company) which is an extension package of MATLAB (registered trademark: MathWorks). Can be executed. FIG. 8 is a diagram conceptually illustrating noise removal by the noise removal unit 16. FIG. 8A shows one of the temporary tile images X before noise removal, and FIG. 8B shows the temporary tile image X after noise removal (hereinafter referred to as temporary tile image X ′).

  In the temporary tile image X shown in FIG. 8A, a noise component exists in the entire image. On the other hand, when noise removal processing is executed, noise components are removed as shown in FIG. However, a “discontinuous region” H occurs at the peripheral edge of the temporary tile image X ′. The discontinuous region H is caused by the fact that the image is interrupted at the periphery of the temporary tile image X during the noise removal processing, and details will be described in comparison with Comparative Example 2 described later. The noise removal unit 16 performs noise removal processing on all the temporary tile images X, and supplies the generated temporary tile images X ′ to the storage unit 18 again for storage.

  The tile image generation unit 17 reads the noise-removed temporary tile image X ′ stored in the storage unit 18 and generates a “tile image”. FIG. 9 is a diagram conceptually illustrating the generation of a tile image by the tile image generation unit 17. FIG. 9A shows the temporary tile image X ′, and FIG. 9B shows the tile image Y generated by the tile image generation unit 17. The tile image generation unit 17 extracts the tile area R1 from the temporary tile image X ′ (that is, removes a part other than the tile area R1) and sets it as the tile image Y. As described above, the discontinuous area H present in the temporary tile image X ′ is not included in the tile area R1 (the tile area R1 is set as such), and thus is not included in the tile image Y. . The tile image generation unit 17 generates tile images Y for all the temporary tile images X ′, supplies them to the storage unit 18, and stores them.

  By the operation from the image acquisition unit 11 to the tile image generation unit 17, the tile image Y generated from each scan image is stored in the storage unit 18. The display unit 19 can read these tile images Y from the tile image generation unit 17 and display them on a display or the like in accordance with an instruction from the user. FIG. 10 shows an image (arranged tile image Y) displayed by the display unit 19.

  As described above, the tile image Y is generated from the scan image G. Since the temporary tile image generation unit 15 divides the scan image G into the temporary tile images X and stores them in the storage unit 18, the memory can be released at this point. As a result, the image acquisition unit 11 can quickly acquire the next scan image G.

  Further, as described above, the temporary tile area setting unit 14 sets the temporary tile area R2 including the tile area R1, and the tile image generation unit 17 extracts only the tile area R1 from the temporary tile image X ′ to obtain the tile image Y. By generating, the discontinuous area H generated by the noise removal processing is not included in the tile area R1. Therefore, when the display unit 19 arranges and displays the tile images Y, it is possible to make the display image not include a discontinuous region.

<Comparison with other noise removal processing>
[Comparative Example 1: When removing noise before tiling a scanned image]
As a comparison, consider a case where noise removal processing is performed on a scanned image before the scanned image is divided into tile images. FIG. 11 is a diagram conceptually illustrating the noise removal method according to the first comparative example.

  FIG. 11A shows a scanned image A1. In this comparative example, this scan image A1 is subjected to noise removal processing to generate a scan image A1 ′ shown in FIG. 11B, and the scan image A1 ′ is further divided to form a tile image A2 shown in FIG. Is generated. Finally, the image (arranged tile image A2) shown in FIG. 11 (d) becomes the display image.

  In this case, since the scan image A1 generally has a large image size (number of pixels), the speed of the noise removal process is extremely slow. For this reason, a very long time is required until the scan image A1 'is generated from the scan image A1. Since the memory can be released only by dividing the scan image A1 ′ into tile images A2 and storing them in the HDD or the like, the next scan image is immediately taken after the image pickup apparatus picks up one scan image. Imaging cannot be performed, and a very long time is required for imaging the entire observation object.

[Comparative Example 2: When removing noise after tiling a scanned image]
Further, as a comparison, a case where noise removal processing is performed on a tile image after the scan image is divided into tile images will be considered. FIG. 12 is a diagram conceptually illustrating the noise removal method according to the second comparative example.

  FIG. 12A shows a scanned image B1. In this comparative example, this scan image B1 is divided to generate a tile image B2 shown in FIG. 12B, and noise removal processing is performed on each tile image B2 to obtain a tile image B2 ′ shown in FIG. Is generated. Finally, the image (arranged tile image B2 ') shown in FIG. 12D becomes the display image.

  In this case, since the scan image B1 is divided into tile images B2 without performing noise removal processing, it does not take time to generate the tile images B2. Therefore, by saving the tile image B2 and releasing the memory, it is possible to capture the next scan image immediately after the image capturing apparatus captures one scan image.

  However, this method has a problem of “discontinuous region”. As described above, when the noise removal process is performed on the tile image B2, the image is interrupted at the periphery of the tile image B2, and therefore information (luminance value and the like) of adjacent pixels is lost. Therefore, an area (discontinuous area J) that does not have continuity with the adjacent tile image B2 'occurs at the periphery of the tile image B2'. That is, as shown in FIG. 12D, in the image displayed on the display unit, the discontinuous region J exists at the boundary of the tile image B2 '.

  This discontinuous area J is a hindrance when the user observes the display image, and in pathological diagnosis, such an image is often further subjected to image processing (such as bright spot counting processing in a fluorescent image). In some cases, it may cause false detection.

[Comparison with Comparative Examples 1 and 2 of the Image Processing System According to the Present Embodiment]
In contrast to Comparative Example 1, in the image processing system 1, the temporary tile image generation unit 15 generates the temporary tile image X from the scan image G and stores the temporary tile image X in the storage unit 18 before executing the noise removal processing. Therefore, unlike the first comparative example, there is no problem that the scan image capturing is delayed because time is required until tile division (the memory is released).

  In contrast to Comparative Example 2, in the image processing apparatus 10, the tile image generation unit 17 sets only the tile region R1 of the temporary tile image X ′ that does not include the discontinuous region H as the tile image Y. Therefore, the discontinuous area H is not included in the tile image Y (which will be the final display image) as in Comparative Example 2, and a display image that is not affected by the discontinuous area H can be generated.

<About scanned images>
The scan image to be processed by the image processing system 1 according to the present embodiment is not particularly limited as long as it is a microscope image, but a fluorescence microscope image captured by a fluorescence microscope is particularly effective. A fluorescence microscope is a microscope that irradiates an observation object stained with a specific staining agent with excitation light and images fluorescence emitted from the observation object.

  In the fluorescence microscope, the brightness of the observation object is smaller than that of a normal microscope (bright field microscope) that targets visible light, and therefore it is necessary to execute exposure by the imaging device for a long time. For this reason, noise is easily generated in the fluorescence microscope image, and it is common in the field of pathological diagnosis using a fluorescence image to perform noise removal processing on the fluorescence microscope image. Therefore, the image processing system 1 according to the present embodiment is particularly effective for fluorescence microscope images.

<Specific configuration of image processing system>
A specific configuration of the image processing system 1 according to the embodiment of the present technology will be described. FIG. 13 shows a fluorescence microscope 200 which is an example of the imaging device 20 of the image processing system 1, and FIG. 14 shows a microscope computer 100 which is an example of the image processing device 10 of the image processing system 1.

  As shown in FIG. 13, the slide glass stock 201 holds a plurality of slide glasses S on which the observation object P is placed. The transfer robot 202 takes out one slide glass S from the slide glass stock 201 and places it on the stage 203.

  The slide glass S is imaged by the macro image capturing camera 205 while being illuminated from the macro image illumination 204 on the stage 203. The macro image capturing camera 205 outputs the captured macro image to the camera control board 206. The camera control board 206 supplies the macro image to the mother board 101 of the microscope computer 100.

  The mother substrate 101 performs an imaging area detection process on the macro image of the slide glass S, and selects an area to be imaged by a microscope. The mother board 101 supplies information on the selected area (hereinafter, area information) to the system control board 207.

  The slide glass S is conveyed by the stage 203 to the imaging range of the microscope. The system control board 207 controls the stage control unit 208 based on the area information, and finely adjusts the stage 203 so that the observation object P of the slide glass S is within the imaging range of the microscope.

  The system control board 207 controls the fluorescent light source 209 to generate fluorescent illumination light. The fluorescent illumination light passes through the excite filter 212 through the light guide 210 and the fluorescent illumination optical system 211 and becomes excitation light. The excitation light is reflected by the dichroic mirror 213, converged by the objective lens 214, and irradiated on the observation object P. In the observation object P, fluorescence is generated by the incident excitation light.

  The fluorescence generated in the observation object P is magnified by the objective lens 214 and travels straight through the dichroic mirror 213. Light having a wavelength other than fluorescence is removed by the emission filter 215, and the image is picked up by the image pickup device 217 (CMOS (Complementary Metal Oxide Semiconductor) or the like) through the imaging lens 216. A fluorescence image (scanned image) captured by the image sensor 217 is output to the camera control board 206.

  Further, the phase difference detection unit 218 may detect the phase difference of fluorescence before imaging and supply the detected fluorescence to the mother substrate 101 of the microscope computer 100. A focus signal that defines the depth of focus is generated from the phase difference of fluorescence by the mother substrate 101, and can be reflected in stage control by the stage control unit 208 via the system control substrate 207.

  As shown in FIG. 14, the scan image captured by the image sensor 217 is supplied to a GPGPU (General-purpose computing on graphics processing unit: GPU capable of general-purpose arithmetic processing) 102 of the microscope computer 100.

  The GPGPU 102 develops the scanned image G (image acquisition unit 11) and supplies it to the mother substrate 101. Further, the GPGPU 102 may perform pixel compression of the scan image G and display the scan image on the display 103. The mother substrate 101 performs stitching of the scanned image G (stitching unit 12), and sets the tile region R1 and the temporary tile region R2 (tile region setting unit 13 and temporary tile region setting unit 14). The mother substrate 101 supplies the scan image G in which the tile area R1 and the temporary tile area R2 are set to the GPGPU 104.

  The GPGPU 104 performs JPEG coding on the scanned image G to generate a temporary tile image X (temporary tile image generation unit 15), and stores it in the HDD 105 (storage unit 18). The mother board 101 reads the temporary tile image X stored in the HDD 105 and executes noise removal processing (noise removal unit 16). The mother board 101 extracts the tile area R1 from the temporary tile image X ′ generated by the noise removal process, and generates the tile image Y (tile image generation unit 17). The mother board 101 stores the tile image Y in the HDD 105. The mother board 101 transmits the generated tile image Y to the server via the network card 106 as necessary.

<Effect of image processing system>
The effect of the image processing system according to the present embodiment will be described. FIG. 15 shows various images. FIG. 15A shows a scanned image that has not been subjected to noise removal processing (200 pixels × 100 pixels), and FIG. 15B shows that noise removal processing has been performed after tiling (equivalent to Comparative Example 2 above). It is a tile image. FIG. 15C is a tile image that has been subjected to noise removal processing by the image processing system 1 according to the present embodiment.

  FIGS. 16 to 19 are graphs showing the luminance distribution in each image shown in FIGS. 16 to 19 show lines perpendicular to the seams of the tiles (excluding FIG. 15A) of each image shown in FIGS. 15A to 15C (white broken lines in FIGS. 15A to 15C). Brightness distribution in (shown). FIG. 16 shows the luminance distribution for red (R), FIG. 17 for green (G), and FIG. 18 for blue (B). In addition, FIG.16 (b), FIG.17 (b), FIG.18 (b) is the graph which expanded the vertical axis | shaft of Fig.16 (a), Fig.17 (a), and Fig.18 (a), respectively.

  Each of the plots in FIGS. 16 to 18 is “before noise removal processing (original) (corresponding to FIG. 15A)” and “noise removal after tiled (tile denoising) (corresponding to FIG. 15B)). "," Noise removal (subtile denoising of this embodiment (corresponding to FIG. 15C)) ".

  As shown in FIGS. 16 to 18, the plot of “tile denoising after tiled” is discontinuous at the tile boundary as compared to “before noise removal (original)”. On the other hand, the plot of “noise removal (subtile denoising of this embodiment)” is not discontinuous, and is close to “before noise removal processing (original)”.

  Therefore, from these plots, compared with the scanned image before the noise removal processing shown in FIG. 15A, the tile images that are subjected to the noise removal processing after tiling shown in FIG. It can be seen that a “seam” (indicated by an arrow in the figure) resulting from the discontinuous region occurs at the boundary. On the other hand, in the tile image shown in FIG. 15C, it can be seen that there is no seam. That is, it can be said that the image processing system 1 according to the present embodiment can prevent generation of discontinuous regions and obtain a highly accurate tile image.

  As described above, the microscope image processing system according to the present embodiment quickly releases the memory by generating a temporary tile image. Further, the image processing system generates a tile image excluding a peripheral portion where a discontinuous region may occur in a temporary tile image subjected to noise removal processing, so that there is no discontinuous region at the boundary of the tile image. A tile image is generated. Therefore, the microscope image processing system according to the present embodiment can generate a microscope image with high speed and high accuracy.

  The present technology is not limited only to the above-described embodiments, and can be changed without departing from the gist of the present technology.

  In addition, this technique can also take the following structures.

(1)
An image acquisition unit for acquiring a scanned image;
A tile area setting section for dividing the scanned image into a plurality of tile areas;
A temporary tile area setting unit that sets a temporary tile area that includes the tile area and is larger than the tile area in the scanned image;
A temporary tile image generation unit that generates the temporary tile image by extracting the scan image for each temporary tile region;
A noise removal unit that performs noise removal processing on the temporary tile image;
An image processing apparatus comprising: a tile image generation unit that extracts the temporary tile image for each tile region and generates a tile image.

(2)
The image processing apparatus according to (1) above,
The image acquisition unit acquires a plurality of continuous scan images,
The image processing apparatus further includes a stitching unit that calculates relative positions of the plurality of scan images.
The tile area setting unit is an image processing apparatus that divides the scanned image into a plurality of tile areas based on the relative position.

(3)
The image processing apparatus according to (1) or (2) above,
The scanned image is an image processing apparatus that is a fluorescence microscope image captured through a fluorescence microscope.

(4)
An imaging device for imaging a scan image of an observation object;
An image acquisition unit that acquires the scan image, a tile region setting unit that divides the scan image into a plurality of tile regions, and a temporary tile region that includes the tile region in the scan image and that is larger than the tile region. A tile area setting unit; a temporary tile image generation unit that generates a temporary tile image by extracting the scan image for each temporary tile region; a noise removal unit that performs noise removal processing on the temporary tile image; and the temporary tile An image processing system comprising: a tile image generation unit that extracts an image for each tile region and generates a tile image.

(5)
The image processing system according to (4) above,
The imaging apparatus is an image processing system that images the observation object via a fluorescence microscope.

(6)
An image acquisition unit for acquiring a scanned image;
A tile area setting section for dividing the scanned image into a plurality of tile areas;
A temporary tile area setting unit that sets a temporary tile area that includes the tile area and is larger than the tile area in the scanned image;
A temporary tile image generation unit that generates the temporary tile image by extracting the scan image for each temporary tile region;
A noise removal unit that performs noise removal processing on the temporary tile image;
An image processing program that causes a computer to function as a tile image generation unit that extracts the temporary tile image for each tile region and generates a tile image.

DESCRIPTION OF SYMBOLS 1 ... Image processing system 10 ... Image processing apparatus 11 ... Image acquisition part 12 ... Stitching part 13 ... Tile area setting part 14 ... Temporary tile area setting part 15 ... Temporary tile image generation part 16 ... Noise removal part 17 ... Tile image generation Unit 20 ... Imaging device

Claims (6)

An image acquisition unit for acquiring a scanned image;
A tile area setting unit that divides the scanned image into a plurality of tile areas;
A temporary tile area setting unit that sets a temporary tile area that includes the tile area and is larger than the tile area in the scanned image;
A temporary tile image generating unit that generates the temporary tile image by extracting the scan image for each temporary tile region;
A noise removing unit that performs a noise removing process on the temporary tile image;
An image processing apparatus comprising: a tile image generation unit that extracts the temporary tile image for each tile region and generates a tile image. The image processing apparatus according to claim 1,
The image acquisition unit acquires a plurality of continuous scan images,
The image processing apparatus further includes a stitching unit that calculates relative positions of the plurality of scan images.
The tile area setting unit is an image processing apparatus that divides the scanned image into a plurality of tile areas based on the relative position. The image processing apparatus according to claim 1,
The image processing apparatus, wherein the scan image is a fluorescence microscope image captured through a fluorescence microscope. An imaging device for imaging a scan image of an observation object;
An image acquisition unit that acquires the scanned image, a tile region setting unit that divides the scanned image into a plurality of tile regions, and a temporary tile region that includes the tile region and is larger than the tile region in the scanned image. A tile area setting unit; a temporary tile image generating unit that generates a temporary tile image by extracting the scan image for each temporary tile region; a noise removing unit that performs noise removal processing on the temporary tile image; and the temporary tile An image processing system comprising: a tile image generation unit that extracts an image for each tile region and generates a tile image. The image processing system according to claim 4,
The imaging apparatus is an image processing system that images the observation object via a fluorescence microscope. An image acquisition unit for acquiring a scanned image;
A tile area setting unit that divides the scanned image into a plurality of tile areas;
A temporary tile area setting unit that sets a temporary tile area that includes the tile area and is larger than the tile area in the scanned image;
A temporary tile image generating unit that generates the temporary tile image by extracting the scan image for each temporary tile region;
A noise removing unit that performs a noise removing process on the temporary tile image;
An image processing program that causes a computer to function as a tile image generation unit that extracts the temporary tile image for each tile region and generates a tile image.